Based on the observations regarding Hand Detection and Tracking, we can conclude that using
YUV Skin Color Segmentation followed by CAMSHIFT algorithm will help in the effective
detection and tracking as the centroid values can easily be obtained by calculating the moments at
each point, later we could combine Hidden Markov Training for further applications. It is better
when compared to Time-Flight Camera where one has to find the bounding box and then use
Iterative Seed Fill algorithm.
For Nose Detection and Tracking system, the results obtained using Edge Detection Techniques
were comparatively better, and it also detects the nose for faces at various angles. The success
rate using this method for frontal faces is about 93% and for non frontal faces it is about 68%.
This shows that this particular method is suitable to be implemented in an automatic 3D face
recognition system.
For the analysis of Eye Detection and Tracking, it would be efficient to implement a method
which would be a combination of texture based and colour based eye detection methods as colour
plays a major role in both texture recognition of facial regions and later finding the accurate
location of eyes using eye maps. Individually the efficiencies of texture based and color based
detection methods are 98.2% and 98.5% respectively, but after incorporating the proposed
method the efficiency could be enhanced to greater extent.